The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted natural language data (token names, descriptions, and symbols) from the TRON blockchain.
Ingestion points: Untrusted data enters the agent context through commands like sun sunpump token get, sun sunpump token ranking, and sun sunpump token search as documented in SKILL.md.
Boundary markers: The instructions lack explicit boundary markers or delimiters to help the agent distinguish between legitimate token metadata and potential instructions embedded within those fields.
Capability inventory: The skill possesses significant capabilities, including executing financial transactions (sun swap, sun sunpump buy/sell) and launching new tokens (sun sunpump launch), which could be targeted by an injection attack.
Sanitization: There is no evidence of sanitization or filtering applied to external content before it is interpolated into the agent's prompts.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of an external command-line tool from a public registry.
Evidence: The README.md and SKILL.md files instruct the user to install @sun-protocol/sun-cli@^1.2.2 globally via npm.
[COMMAND_EXECUTION]: The skill relies on shell command execution to perform its primary functions.
Evidence: All core features (trading, launching, querying) are implemented as wrappers around the sun executable installed via the sun-cli package.

SunPump Meme Token Toolkit